While alcohol abuse among the elderly is an increasing problem, there is little information on their patterns of alcohol use and abuse. Information from longitudinal studies of alcohol consumption and abuse may provide the most effective means of improving prediction of alcohol problems in later life. The Normative Aging Study (NAS) offers a unique database for examining change and stability in drinking patterns over a substantial part of the adult lifespan. In 1973 and 1982, over 2000 men initially aged 28 to 89 responded to one or two detailed mail surveys regarding alcohol use. We also administered the alcohol survey to 1,709 men in the fall of 1991, in order to maintain the nine year interval between data assessments. The proposed project consists of four studies whose overall purpose is the description and prediction of alcohol abuse in later life. In Study 1, we will examine patterns of change and stability in alcohol consumption and turnover in heavy drinking. Analysis of the first two data points indicated considerable change in drinking behavior over time, with 39% changing consumption categories, 44% moving from heavy drinking to a lighter drinking category, and 45% moving from problem to nonproblem drinking. The third point allows us to assess further who develops and who recovers from alcohol abuse in middle and later life. Study 2 will use measures from four domains - family history, social, personal, and situational - to assess the risk for heavy drinking and for alcohol problems in later life. Using appropriate methods, we will examine the contribution of measures from each domain to the sensitivity and specificity of a classification model, and identify misclassified men, e.g. men who score high on the MAC scale but who do not report drinking problems. In Study 3 we will interview a subsample of 250 individuals, oversampling those who are misclassified with regard to alcohol abuse, to identify ways of reducing the misclassification rate. The interviews will also focus on proximal reasons for alcohol abuse, strategies for coping with stress, and on age-specific alcohol problems. Based on the interviews, Study 4 will develop a questionnaire to be administered to the remaining men in the NAS to validate both our classification model and the new measures derived from the interviews. These studies will provide information that will improve our understanding of the development and cessation of alcohol abuse in later life.